In general, electrical connectors consist of two components, a receptacle and a plug. The receptacle is the compliant part of the connector. That is, the receptacle is fashioned in such a way that it provides compliance (or “springiness”), either through the use of a springy metal such as a Beryllium-Copper (Be—Cu) alloy or some other means. The plug then forms the non-compliant part of the connector.
Connectors are used in a variety of applications where electrical coupling between components, e.g., integrated circuits, circuit boards, etc., is desired. However, connectors for high speed interfaces are required to present controlled impedance interconnections. The interface between a Rambus DRAM (RDRAM®) and a Rambus Channel is an example of a high speed interface that requires a connector having particular electrical and physical characteristics.
Since the early 1970s, the essential characteristics of a DRAM interface have remained as a separate data bus and a multiplexed address bus. However, a recent architecture pioneered by Rambus, Inc. provides a new, high bandwidth DRAM interface. Originally, the Rambus Channel, the heart of the new DRAM interface, comprised a byte wide, 500 or 533 Mbytes/sec. bi-directional bus connecting a memory controller with a collection of RDRAMs®. Among the many innovative features of the Rambus Channel and of the RDRAM® is the use of vertically or horizontally mounted RDRAMs® and a physically constrained, bi-directional bus using terminated surface-trace transmission lines on a circuit board. The physical and electrical properties of both the RDRAMs® and bus on which they are placed are rigidly defined because high frequency operation relies on the careful physical design of both the printed circuit board and the high speed components. Originally, RDRAMs® were specified to include a 32-pin package, either a surface horizontal package (SHP) or a surface vertical package (SVP).
Electrical connectors of the past have generally been unsuitable for use in high speed bus applications such as may be found with the Rambus Channel. For example, as shown in FIG. 1, electrical connectors of the past have employed compliant contact elements 2 to receive semiconductor devices and/or circuit boards to provide electrical coupling to a circuit on a substrate 4 (e.g., a motherboard). The electrical connectors may be contained within housings 6 adapted to receive the semiconductor device or circuit board and are electrically coupled to circuit elements on the motherboard through a solder connection 8. Such a connector thus requires a number of surface mount contacts (e.g., solder contacts 8) between the contact elements 2 and the substrate 4.
Such a connector is not suitable for use in a high speed electrical bus because the contact elements 2 are individually soldered to circuit elements (e.g., electrical traces) on the substrate 4, and because the resulting solder joints 8 are generally not accessible for inspection and repair. High speed bus design dictates that the electrical signal path from device to device be kept at a minimum. Further, electrical contacts on each device should be concentrated into a small area. Together, these requirements lead to a high density area array of separable contacts, whose corresponding solder joints are made inaccessible due to interference from adjacent contacts and/or the contact housing. Except for special “ball grid array” soldering techniques, surface mount solder joints are generally required to be accessible for inspection and repair. Because connectors such as that illustrated in FIG. 1 are incapable of meeting these requirements, they are unsuitable for use in high speed bus applications.